Shopify's Meta AI channel: product image variants from Sume
Shopify now lists Meta as an AI channel and shares products through Shopify Catalog. Here is what Sume can make for those listings: image variants and a video.

Shopify's changelog says Meta now appears alongside other AI channels in Agentic Storefronts, and that products are shared with Meta by default through Shopify Catalog. Sume does not connect to that channel; what it does is make the product media you put in your Shopify catalog: a per-model range for images per request (1 to 4 in the docs example), in a stated list of aspect ratios, plus a video through a Format.
Shopify's statements are from its changelog entry, read 2026-10-01. Sume facts are from the image generation and bulk runs docs.
What did Shopify announce?
The entry says Meta is now an AI channel in your admin, products are shared with Meta by default through Shopify Catalog, and you can manage access for Catalog and direct checkout and view performance for Meta in your agentic settings. It does not list image sizes or video requirements, so use Shopify's and Meta's own product media specs for those.
How many image variants can one request make?
The image models endpoint reports n as a range per model. The docs example for bytedance-seed/seedream-4.5 shows n from 1 to 4, input_references from 0 to 10, and output_format of png, jpeg or webp. Read supported_parameters for the model you pick, because a model only accepts the values its catalog descriptors list.
| Control | What the docs say |
|---|---|
n | Range per model; 1 to 4 in the docs example |
aspect_ratio | Normalized ratio such as 1:1, 16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 4:5; auto lets the provider choose |
4:5 | Instagram portrait (1080 x 1350), not 4:3 |
| Exact 1080 x 1350 | A documented post-step via job target_pixels, not the native size |
Which aspect ratios should a catalog set include?
The docs list 1:1 and name 4:5 as the Instagram portrait shape. Generate the same product in a few ratios by changing aspect_ratio only, and keep the prompt and input_references the same so the product stays recognizable. The post facebook catalog extra product angles shows the reference approach.
How do I do this for a whole catalog?
For video, a first-party Format can be run per product, and a bulk request queues up to 100 Format runs with a concurrency window. Poll GET /v1/format-run-queues/{id} and read each child at GET /v1/format-runs/{run_id}. See product video for Shopify Catalog for the video side.
Sources
Related posts
More in Integrations
- Slack incoming webhook 1/sec: when 100 Sume jobs finish together
Slack incoming webhooks allow about 1 message per second. When 100 Sume jobs finish at once, store each webhook, then post to Slack from a paced queue.
- Snapchat Ads MCP and hosted Sume tools in one agent
Snap's Ads MCP server answers campaign questions and is read-only at launch. Add Sume's hosted MCP in the same agent to turn the findings into creative.
- Tavus PAL MCP connectors vs Sume hosted MCP tools and gates
Tavus MCP connectors let a PAL call third-party MCP servers. Here is what Sume's hosted MCP at mcp.sume.com exposes and the gates a paid call needs.
- Telegram Bot API 10.3 ephemeral sendVideo with a Sume clip
Bot API 10.3 adds ephemeral_message_parameters to sendVideo. Send a finished Sume artifact URL from your webhook handler; steps, retries and limits.
Written by Sume